Integration by Parts Examples: Choose u and Check Each Answer

Consider . The power rule cannot integrate the whole product, and integrating alone leaves the behind. Integration by parts gives the product a useful direction: differentiate the part that becomes simpler, integrate the part that stays manageable, and then handle the smaller integral that remains.
This lesson works through integration by parts examples, from a single product to repeated use and definite bounds. If you want the meaning of an integral before the technique, start with our integral guide. The derivative guide also helps because differentiating your answer is the best check here.
Where the formula comes from
The product rule says . Integrate both sides and move one term across. This gives the integration by parts formula described in OpenStax Calculus, Volume 2:
Here is the expression you choose to differentiate, and is what you choose to integrate. After finding and , the new integral should be easier than the original one. If it is worse, reconsider the split. The formula is an algebraic consequence of the product rule, not a recipe that guarantees every split will help.
For an indefinite integral, remember the constant at the end. The choice of an antiderivative for can absorb a constant, but it is simpler to pick one convenient and add to the final answer.
Choose u and dv with a purpose
The useful question is: “Which factor shrinks when differentiated?” A polynomial such as becomes , then , then zero. A logarithm becomes , which often combines well with the rest. Meanwhile, exponentials and sine or cosine remain easy to integrate repeatedly.
The familiar LIATE mnemonic orders logarithmic, inverse trigonometric, algebraic, trigonometric, and exponential factors as candidates for . OpenStax presents it as a heuristic, not a theorem. Check the resulting instead of trusting the first letter you see. A different split may be shorter, and some products are better handled by substitution.
For example, contains an inner function and its derivative . Set . Then and the integral is . Using integration by parts would turn a simple composition into unnecessary work. In contrast, has no such direct inner-derivative match, while differentiating removes it.
Write a small two-column table before calculating: , on one side, , on the other. Then ask whether multiplying and creates a simpler integral. This check prevents the common mistake of choosing the easy-to-integrate factor as merely because it looks familiar.
Example 1: a linear factor times an exponential
Find . Choose so . Choose , giving . The factor matters: differentiating produces an extra two.
Differentiate the factored answer. The product rule gives , which simplifies to . That restores the integrand exactly. If the check had produced , the likely error would be omitting the when integrating .
The choice also has a visible stopping point: differentiating leaves a constant, so the remaining integral contains only the exponential. That is what “simpler” means in this example.
Example 2: integrate a logarithm
Find for . There is no written product, but you can supply a factor of one: . Choose and . Then and .
The check is short: . The domain lets us write and without crossing zero. The more general antiderivative works on intervals that do not include zero, but the stated problem stays on the positive side.
If you chose and , you would need to know the original integral just to find . That circular split explains why the logarithm belongs in here.
Example 3: repeat the method until the polynomial disappears
Find . Differentiating reduces its degree, while integrating sine remains straightforward. First choose and , so and :
The remaining integral still contains a product. Let and apply parts again: , , , . Then
Substitute that result into :
Now differentiate term by term. The derivative of is . The derivative of is ; that of is . All extra terms cancel, leaving . This cancellation is a strong check on both the sign from integrating sine and the factor of two from differentiating .
Repeated integration by parts is useful when each application makes a factor smaller. If the new integral resembles the original instead, it may still be solvable by collecting the original integral on one side, but do not keep applying the formula without a clear end point.
Example 4: keep the bounds in a definite integral
Evaluate . The factor is easy to integrate, but differentiating produces . Choose and . Then and . The definite form of integration by parts is .
Polynomial division gives . Its integral from zero to one is . Therefore
This positive result is plausible because is nonnegative on . As an independent algebra check, an antiderivative is ; evaluating at one and zero also gives . For definite integrals, carry the same endpoints through the new integral. There is no in the final number.
Three practice problems with solutions
1. A polynomial and an exponential. Find .
Take and . Then and , so
Differentiate : it gives . The result matches the original integrand.
2. A logarithm with a coefficient. Find for .
Choose and . Then and :
Differentiation gives . Keeping the two terms visible makes the check easier to follow.
3. A definite integral. Evaluate .
Take and , giving and . Apply the bounds after the parts step:
Both and are nonnegative on this interval, and is positive. Differentiating gives , which checks the antiderivative used to evaluate the bounds.
To practise the underlying antiderivative choices, use Math Zen's integrals topic. For a later application of repeated integration methods, see the Laplace transforms guide.
Common Questions
- How do I decide between substitution and integration by parts?
- Look first for an inner function accompanied by its derivative, as in 2x cos(x²). Substitution is direct there. If you have a product whose differentiated factor becomes simpler, as in x e^(2x), try parts. Some integrals need both techniques in sequence.
- Must I follow LIATE exactly?
- No. LIATE suggests a first choice for u. Judge that choice by the new integral of v du. If it is harder than the original or leads in a circle, choose again.
- Can I check every answer by differentiating?
- For an indefinite integral, differentiating the proposed antiderivative must recover the integrand on the stated domain. For a definite integral, check the antiderivative first and then evaluate its endpoint difference. A numerical estimate can also catch an implausible sign or size.
- Why is there no +C on a definite integral answer?
- A definite integral is a number. If you use an antiderivative F + C, the constants cancel in (F(b) + C) − (F(a) + C).


